The dangerous PDF to Excel conversion is not the one that fails. A failed conversion is obvious: you open the file, the columns are shredded, you try something else. The dangerous one is the conversion that opens clean, looks right, and is quietly missing four rows from the bottom of page three.
Nobody catches that by scrolling. Below is the check accountants and analysts actually run, ordered so the fastest tests that catch the worst errors come first. It takes about ten minutes on a typical document and it is the difference between using converted data and trusting it.
How do I know if my PDF to Excel conversion is accurate?
Check three things, in this order: that the row count matches the source, that a totaled column ties to the total printed on the PDF, and that the amounts are stored as numbers rather than text. Those three tests catch the overwhelming majority of real conversion errors. Everything else is a spot check on individual cells, which is slower and finds less.
Check 1: does the row count match?
Count the records in the source and count the rows in the spreadsheet. If the PDF says 143 invoices and the sheet has 139, four rows went somewhere, and no amount of cell-level inspection will tell you which four.
In Excel, =COUNTA(A2:A5000) on a column that is always populated gives you the record count fast. Compare it to the count printed on the document, if it prints one. Many financial reports helpfully state "143 records" or number their rows, and that line is the cheapest audit you will ever run.
Rows disappear for two reasons. A table that breaks across pages can lose the rows nearest the page boundary, especially where a footer or a repeated header interrupts the grid. And a row whose first cell is blank sometimes gets read as a continuation of the row above, silently merging two records into one. Both are page-break problems, which is why the check matters most on long documents.
Check 2: do the totals tie?
This is the single strongest test, because it is the one that catches errors you did not think to look for. Sum the main amount column in Excel and compare it to the total printed at the bottom of the PDF.
| Result | What it means |
|---|---|
| Sums match exactly | Row count and amounts are almost certainly correct. |
| Excel total is lower | Rows were dropped, or some amounts are text and not being summed. |
| Excel total is higher | A subtotal or repeated header row got captured as data. |
| Excel shows 0 | The whole column is text. Nothing is being added at all. |
| Off by a round factor | A decimal or thousands separator was misread, common in OCR. |
The "Excel total is higher" case deserves particular attention. Financial PDFs are full of subtotal lines, and a converter that treats a subtotal as an ordinary record double counts everything above it. If your sum is roughly double the printed total, that is what happened.
Why are my numbers wrong after converting a PDF to Excel?
Usually because they are not numbers. Amounts converted out of a PDF frequently land as text, so Excel displays them, left-aligns them, and refuses to add them. Select the column and look at the status bar: if there is a count but no sum, they are text. Select the column, choose Data, then Text to Columns, and click Finish, and they become real numbers.
Currency symbols, parentheses for negatives, and trailing spaces all cause this, and each has its own quirk. We cover the specific fixes in guides on numbers stored as text, currency symbols, and negative numbers in parentheses. The tell for all of them is the same: the column will not sum.
Check 3: are the columns still aligned?
Column drift is the error most likely to survive every other check, because the totals can still tie while individual rows carry the wrong values. It happens on wide tables, where a blank cell in one row shifts everything to its right by one column.
Sort the sheet by a column that should have a predictable shape and see what falls out of place. Sort by a date column and any row where a name landed in the date field jumps to the top or bottom immediately. Sort by an amount column and text values cluster at one end. Filter a status column and read the distinct values; a stray tenant name in a status field is instantly visible in a filter dropdown that should only ever contain "Paid" and "Open".
Then check the first row, the last row, and the row at each page boundary against the PDF, cell for cell. Page boundaries are where the structure breaks, so those are the rows worth reading by eye.
How do I check a scanned PDF conversion?
Scans need one extra pass, because OCR does not fail loudly. It guesses, and a confident wrong guess reads exactly like a right one. The characters it confuses are predictable: 5 and S, 0 and O, 1 and l and I, 8 and B, and rn read as m.
Two quick screens. First, sort every numeric column ascending and descending and look at the extremes. An OCR error that turns 1,200 into 11,200 or 120 shows up as an outlier at one end. Second, run =ISTEXT(cell) down a column that should be entirely numeric, or just filter it: any value containing a letter is an OCR misread, because a proper amount never does.
Scan quality drives all of this. At 300 DPI, results are reliable. Below 200 DPI, plan on checking every page. If the source documents are receipts and small-format paper rather than clean reports, that is a different extraction problem, and reading the line items off a receipt is worth handling with a tool built for it. A better OCR conversion also removes most of this work at the source.
What is a good accuracy rate for PDF to Excel conversion?
Be careful with published accuracy percentages, because vendors define them differently and rarely say how. A tool claiming 99 percent character accuracy can still misplace one number in a hundred, which on a 500 row general ledger is five wrong figures, any one of which can throw a reconciliation.
The practical standard is not a percentage. It is whether the totals tie and the row count matches. A conversion where the sum agrees with the printed total and no records are missing is usable, whatever the vendor's marketing says. One where the totals disagree by a dollar is not usable until you know why, because a dollar and a thousand dollars are the same kind of error with different luck.
Do I need to check every page?
No, and doing so defeats the point of converting. The checks above are designed to make a whole-document error visible without reading the whole document. Row count catches dropped records anywhere in the file. A totals tie-out catches a wrong amount anywhere in the column. Sorting catches column drift anywhere in the sheet.
Read individual pages only where those tests point you: the page where the count went wrong, the rows at page boundaries, and the outliers a sort surfaced. On a document you convert every month from the same source, run the full check the first time, then just the totals tie-out afterward.
A ten minute checklist
- Count rows in the sheet, compare to the record count in the PDF.
- Sum the main amount column, compare to the printed total.
- If it shows zero or comes up short, fix text-formatted numbers, then sum again.
- Scan for captured subtotal and repeated header rows, and delete them.
- Sort by date, then by amount, and look at both extremes for values in the wrong column.
- Read the first row, the last row, and each page boundary row against the source.
- On scans, filter numeric columns for anything containing a letter.
- Save the raw converted file untouched before you clean anything.
That last step matters more than it sounds. Once you start deleting rows and reformatting columns, you lose the ability to tell a conversion error from a cleanup error. Keep the raw output on its own tab, work on a copy, and when a number looks wrong two weeks later you will know which one to blame. The full cleanup sequence picks up exactly where this check leaves off.
Most of these errors are conversion quality problems, not Excel problems. A converter that reads table structure properly, keeps amounts numeric, and handles page breaks without repeating headers gives you a file that passes this check on the first try. That is what an accurate PDF to Excel converter is actually selling: not a faster download, but a shorter checklist.